Bug: Feishu channel fails to connect - axios timeout in gateway process

[median glen]

OS: Windows 10, Node v24.11.1, OpenClaw 2026.4.11
channels.feishu configured with websocket mode
Gateway process (standalone node, PID separate from Electron) gets AxiosError: timeout of 60000ms exceeded on every request to https://open.feishu.cn/open-apis/auth/v3/tenant_access_token/internal
Same request from a child process (via exec tool node script.js) succeeds in <200ms
web_fetch tool (also in gateway) can reach open.feishu.cn fine
Tried: multiple restarts, different app credentials, increased httpTimeoutMs to 60000
No proxy, no firewall, DNS resolves in 12ms

[young spruce]

To help others find answers, you can mark your question as solved via Right click solution message -> Apps -> ✅ Mark Solution

[spare fjord]

I checked the Feishu code path. This is very likely not credentials, DNS, or normal network reachability.

What matters here: Feishu channel API calls go through @larksuiteoapi/node-sdk → bundled axios. web_fetch uses a different Gateway HTTP path, and your child-process test is also not the same long-lived Gateway process state. So those passing only prove the host can reach Feishu; they don’t clear the Lark SDK/axios-in-Gateway path.

This matches existing reports pretty closely:

• https://github.com/openclaw/openclaw/issues/61971
• https://github.com/openclaw/openclaw/issues/72721

My best next moves, in order:

1. Try Node 22 LTS for the Gateway, not Node 24. OpenClaw requires Node >=22, but Node 24 + axios/Lark SDK in the Gateway process is a common-looking part of these reports. If it disappears on Node 22, that’s a very useful regression signal.

2. If you need Feishu working right now, use the known workaround: switch Feishu to webhook mode instead of websocket mode. Docs: https://docs.openclaw.ai/channels/feishu

3. Add your exact repro details to the GitHub issue rather than treating this as local config. Especially useful details:

   • Windows 10
   • Node v24.11.1